Epileptic seizure localized by whole head MEG
S. M. Bowyer, K. Mason, B. J. Smith, and G. L. Barkley
Henry Ford Hospital, Detroit, Michigan USA
1 Introduction
Magnetoencephalography (MEG) is
currently used clinically for presurgical
localization of epileptic tissue, based on signals
from interictal spikes, using single equivalent
current dipole (ECD) modeling [1,2,3,4], There
has been a long-standing question “Do interictal
spikes co-localize to the same area as epileptic
seizures?”
Minassin, et al. found that the interictal
MEG localization of spikes corresponded to
ictal zones mapped by the ECoG in ten of
eleven patients [5]. MEG was able to map
interictal activity equally well as intracranial
electrodes (ECoG). This suggests that the
localizing information obtained by the invasive
intracranial monitoring may also be available by
noninvasive MEG.
Mappings of epileptic seizures by MEG
are rare since patient movement typifies most
seizures, and localizing brain activity after the
patient moved is not accurate. In one study,
performed by Ko et al. [6], in which an epileptic
seizure was monitored by MEG, the data was
compared to EEG localizations. The MEG data
localized the active source more mesial in the
temporal lobe than the EEG. The mean
difference in localization between MEG and
EEG interictal spikes was 2.1 cm (patient 1) and
3.8 cm (patient 2). The mean difference in
localization between the ictal and the interictal
data from EEG was 3.5 cm (patient 1), whereas
the mean difference in localization between the
ictal and interictal data from MEG was 1.8 cm,
(patient 2). That study suggests that the MEG
may be more reliable in comparing the interictal
spikes with the ictal spikes.
We report a case study of localization of
MEG data from both interictal spikes and an
epileptic seizure captured by MEG in the same
subject.
2 Methods
2.1 Patient study
A male patient (27 years old) with
complex partial and secondarily generalized
seizures was monitored with 148 channel
Neuromagnetometer (4D Neuroimaging
Magnes WH2500) and 21 channels of EEG.
This patient has persistent intractable
localization-related epilepsy despite two
previous left frontal lobe resections.
The patient changed into a hospital
gown and removed all metal articles from his
body, except for dental work, which was
adequately demagnetized with a commercial
videotape eraser. Three small electrode coils,
used to transmit subject location information to
the neuromagnetometer probe were taped to the
forehead with two-sided tape. Disposable ear
molds of the correct size were placed in the ears
and an additional localization coil was attached
to each ear mold. The EEG electrodes were
applied with collodion adhesive using the
International 10-20 system of measurement.
Impedances of all electrodes were below 5000
ohms. The montage used for recording during
the MEG study was a P z reference montage.
The subject lay comfortably on the bed
inside of the Magnetically Shielded Room
(MSR), and automatic probe position routines
were used to locate the head with respect to the
neuromagnetometer detector coils. The
neuromagnetometer helmet containing the
detector array was then placed over the patient’s
head, in close proximity to most of the cortical
surface. He was instructed to keep his head as
still as possible. His face was visible via video
camera image and there was intercom
communication available between the
technologist and the patient in the shielded
room.
Results
2.2 Data Collection
Parameters for both MEG and EEG
recordings were: low pass filter - 100 Hz; high
pass filter - 0.1 Hz; data was digitized at 290.64
samples per second.
Two 10 minute and one 5 minute
continuous acquisitions were recorded. Visual
inspection of the patient’s face and of the
MEG/EEG real time recording was done.
During the 5-minute acquisition, a seven-second
period of seizure activity was recorded by MEG
and EEG. The patient demonstrated one of his
typical partial seizures characterized by a
staring spell with eyes wide open but with no
body movement. Somatosensory evoked field
studies were also recorded.
3
The single ECD technique localized the
source of activity for both interictal spikes and
seizure onset. Interictal spikes were selected
from the data prior to the epileptic seizure.
Waveforms in Figure 1 show the start of the 5
Hz activity which is the onset of the seizure at
93.56 seconds. Both interictal and ictal sources
were localized in the left frontal region,
approximately 2.2 cm apart as seen in Figure 2.
The parameters for both the seizure and
interictal spikes had similar values and had high
correlations and confidence regions (CR) under
2 cm 3 . RMS values were twice the Q values
and over 400 fTesla. Table 1 lists the dipole fit
parameters for early latencies of the epileptic
seizure and representative interictal spikes.
2.3 Data Analysis
Single ECD software [1,2] was used to
localize the source of activity for both interictal
spikes and the seizure onset. Waveforms were
inspected visually after data was filtered with a
bandpass of 3-100 Hz and a notch filter at 60
Hz. Selected interictal spikes and spikes
occurring during the seizure were mapped using
a single equivalent dipole model. A single
dipole was selected to represent each sharp
wave. The dipole selection criteria [4]
included:
1) Correlation coefficient (R) of 0.98 or
better
2) Root Mean Square (RMS) value of
waveforms across all channels of 400 IT
or more
3) Dipole moment (Q) generally of less
than 400 nAm
4) Confidence region (CR) of less the 3cm ,
'J
preferably less than 1cm .
JUCJUi rA aaJLju
. r wwwvI-a
50mV
95
Seconds
Figure 1: The MEG and EEG wave forms for
epileptic seizure. Red arrow denotes onset of
seizure activity.
In general, the ECD was selected from the
initial onset of the spike waveform up to the
point of maximum amplitude of the spike. The
dipole calculation was performed using 64
magnetometer channels which were chosen to
best represent the contour plot of the magnetic
field.
The epileptic seizure began with an initial sharp
wave arising in the left precentral gyrus (Fig
2B). The next several sharp waves arose
anteriorly towards the surgical margin from the
previous left frontal lobe resection where the
sharp waves and spikes from the seizure
clustered in tight formation. The center of
activity for the interictal spikes was also located
in the anterior portion of the left inferior frontal
gyrus (Fig. 2). Both ictal and interictal activity
co-located in the left inferior frontal gyrus, but
centers of activity were approximately 2.2 cm
apart. The source of the seizure activity was
b) Interictal
c) Interictal
e) Ictal
f) Ictal
a) Interictal
d) Ictal
Figure 2: Interictal spike localization (yellow
triangles): a, axial; b, coronal; c, sagittal. Ictal
seizure localization (red squares): d, axial; e,
coronal; f, sagittal.
more focal than that of interictal spiking and
located more mesial to the surface of the cortex
along the edge of previously resected cortical
tissue. The edges of the previously resected
tissue are seen in the MRI scans. Multiple
source analysis also located the source of
activity in this same region [7], The MRI scans
on the left side of Figure 2 display the interictal
localizations; the scans on the right of Figure 2
display the seizure localizations.
Table 1: The dipole fit parameters for interictal
epileptic spikes.
Interictal
Latency
RMS
fFesla
GoF
Corr.
Q
iiArn
CR
3
cm
166.93
98.73
0.93
0.97
216.93
0.59
174.33
80.49
0.96
0.99
132.93
1.03
505.65
48.73
0.95
0.98
91.61
1.83
512.30
58.46
0.97
0.98
117.64
0.45
Table 2: The dipole fit parameters for ictal
epileptic spikes.
Ictal
Latency
RMS
fFesla
GoF
Corr.
Q
nAm
CR
3
cm
93.56
738.93
0.87
0.89
193.90
0.91
93.68
697.59
0.94
0.93
220.47
0.74
93.79
1073.10
0.94
0.96
316.01
0.45
93.85
799.68
0.91
0.97
173.35
1.12
93.92
678.33
0.97
0.96
243.67
1.25
93.95
1474.80
0.95
0.97
335.20
0.18
94.12
1531.20
0.94
0.93
597.75
0.18
94.18
946.93
0.96
0.96
363.16
0.41
4 Discussion
The question of whether interictal spikes
should be used as the basis for determining the
areas of resecting cortical tissue is unresolved.
In the present case, the zone of ictal onset was
smaller than the zone of interictal activity. The
seizure activity was at the edge of the
previously resected tissue, more mesial to the
cortical surface than the interictal spikes. As
more MEG systems come into use, the
likelihood of co-localization of the epileptic
tissue for seizure and interictal spikes will
increase.
References
1. Ebersole JS, “Magnetoencephalography/
magnet source imaging in the assessment of
patients with Epilepsy”, Epilepsia 38 (Suppl
4)S1-S5, 1997.
2. Wheless JW, Willmore LJ, Breier JI Kataki
M, Smith JR, King DW, Meador KJ, Park
YD, Loring DW, Clifton GL, Baumgartner
J, Thomas AB, Constantinou JEC,
Papanicolaou AC: A comparison of
magnetoencephalography, MRI, and V-EEG
in patients evaluated for epilepsy surgery.
Epilepsia 40: 931-941, 1999.
3. Ebersole JS “A Comparison of
Magnetoencephalography, MRI, and V-
EEG in patients evaluated for Epilepsy
Surgery” Comprehensive Epileptology,
Lippincott-Raven, Philadephia, 919-936,
1997.
4. S. M. Bowyer, K. Mason, N. Tepley and G.
L. Barkley, “Parameters for Clinical
Evaluation of Interictal Epileptic Spikes”,
this volume.
5. Minassian BA, Ostubo H, Weiss S, Elliot I,
Rutka JT, Snead III OC, “Magneto¬
encephalography localization in pediatric
epilepsy surgery: Comparison with invasive
intracranial electroencephalography, Annals
of Neurology 46: 627-633, 1999.
6. Ko DY, Kufta C, Scaffidi D, Sato S. “
Source localization determined by magne¬
toencephalography and electroe¬
ncephalography in temporal lobe epilepsy:
comparison with electrocorticography:
Technical case report.” Neurosurgery 42:
414-422, 1998.
7. Moran JE, Barkley GL, Tepley N, “Two
Dimensional Inverse Imaging (2DII) of
Epileptic Seizures”, this volume.